
import numpy as np
import pandas as pd
import plotly.express as px
import seaborn as sns
import matplotlib.pyplot as plt


df = pd.read_csv('House Price Prediction in India.csv')
df.head(10)


df[['BHK_NO.', 'BHK_OR_RK']]


df['BHK_OR_RK'].value_counts()

BHK    29427
RK        24
Name: BHK_OR_RK, dtype: int64


df.describe()


df['BHK_NO.'].unique()

array([ 2,  3,  1,  4,  5,  6, 12,  8, 20, 10,  7,  9, 13, 17, 15, 11],
      dtype=int64)


df.dtypes

POSTED_BY                 object
UNDER_CONSTRUCTION         int64
RERA                       int64
BHK_NO.                    int64
BHK_OR_RK                 object
SQUARE_FT                float64
READY_TO_MOVE              int64
RESALE                     int64
ADDRESS                   object
LONGITUDE                float64
LATITUDE                 float64
TARGET(PRICE_IN_LACS)    float64
dtype: object


plt.figure(figsize = (15, 6))
sns.heatmap(data = df.corr(), annot = True, cmap = 'RdYlGn')

<AxesSubplot:>


fig = px.bar(x=df["BHK_NO."].unique(), y=df["BHK_NO."].value_counts())
fig.show()


plt.figure(figsize = (15, 6))
sns.barplot(data = df, x = 'RERA', y = 'TARGET(PRICE_IN_LACS)')

<AxesSubplot:xlabel='RERA', ylabel='TARGET(PRICE_IN_LACS)'>


plt.figure(figsize = (15, 6))
sns.barplot(data = df, x = 'POSTED_BY', y = 'TARGET(PRICE_IN_LACS)')

<AxesSubplot:xlabel='POSTED_BY', ylabel='TARGET(PRICE_IN_LACS)'>


plt.figure(figsize = (15, 6))
sns.barplot(data = df, x = 'UNDER_CONSTRUCTION', y = 'TARGET(PRICE_IN_LACS)')

<AxesSubplot:xlabel='UNDER_CONSTRUCTION', ylabel='TARGET(PRICE_IN_LACS)'>


plt.figure(figsize = (15, 6))
sns.barplot(data = df, x = 'BHK_NO.', y = 'TARGET(PRICE_IN_LACS)')

<AxesSubplot:xlabel='BHK_NO.', ylabel='TARGET(PRICE_IN_LACS)'>


df = df.drop(['BHK_OR_RK', 'ADDRESS', 'LATITUDE', 'LONGITUDE'], axis = 1)
df.head()


df = pd.get_dummies(df)


df.columns

Index(['UNDER_CONSTRUCTION', 'RERA', 'BHK_NO.', 'SQUARE_FT', 'READY_TO_MOVE',
       'RESALE', 'TARGET(PRICE_IN_LACS)', 'POSTED_BY_Builder',
       'POSTED_BY_Dealer', 'POSTED_BY_Owner'],
      dtype='object')


df.head()


df = df.drop(['POSTED_BY_Builder'], axis = 1)
df.head()


X = df.drop(columns = ['TARGET(PRICE_IN_LACS)'])
y = df['TARGET(PRICE_IN_LACS)']

X

y

0        55.0
1        51.0
2        43.0
3        62.5
4        60.5
         ... 
29446    45.0
29447    16.0
29448    27.1
29449    67.0
29450    27.8
Name: TARGET(PRICE_IN_LACS), Length: 29451, dtype: float64


from sklearn.preprocessing import StandardScaler
sc = StandardScaler()
X = sc.fit_transform(X)

X

array([[-0.46813431, -0.68271456, -0.44623962, ...,  0.27523994,
        -1.28022595,  1.33968012],
       [-0.46813431, -0.68271456, -0.44623962, ...,  0.27523994,
         0.78111211, -0.74644684],
       [-0.46813431, -0.68271456, -0.44623962, ...,  0.27523994,
        -1.28022595,  1.33968012],
       ...,
       [-0.46813431, -0.68271456, -0.44623962, ...,  0.27523994,
         0.78111211, -0.74644684],
       [-0.46813431, -0.68271456, -0.44623962, ...,  0.27523994,
        -1.28022595,  1.33968012],
       [-0.46813431,  1.464741  , -0.44623962, ...,  0.27523994,
         0.78111211, -0.74644684]])


X.shape

(29451, 8)


y.shape

(29451,)


from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)


X_train

array([[-0.46813431,  1.464741  , -0.44623962, ...,  0.27523994,
         0.78111211, -0.74644684],
       [ 2.13613911,  1.464741  , -0.44623962, ...,  0.27523994,
         0.78111211, -0.74644684],
       [ 2.13613911,  1.464741  , -0.44623962, ..., -3.6331937 ,
         0.78111211, -0.74644684],
       ...,
       [-0.46813431, -0.68271456,  1.82887558, ...,  0.27523994,
        -1.28022595,  1.33968012],
       [-0.46813431,  1.464741  , -0.44623962, ...,  0.27523994,
        -1.28022595,  1.33968012],
       [ 2.13613911,  1.464741  ,  0.69131798, ...,  0.27523994,
         0.78111211, -0.74644684]])


from sklearn.linear_model import LinearRegression
regressor = LinearRegression()
regressor.fit(X_train, y_train)

LinearRegression()


y_pred = regressor.predict(X_test)


from sklearn.metrics import r2_score, mean_squared_error, mean_absolute_error

R2_Score = r2_score(y_test, y_pred)
Mean_Absolute_Error = mean_absolute_error(y_test, y_pred)
Mean_Square_Error = mean_squared_error(y_test, y_pred)
Root_Mean_Square_Error = np.sqrt(mean_squared_error(y_test, y_pred))

results = pd.DataFrame([['Multiple Linear Regression', R2_Score, Mean_Absolute_Error, Mean_Square_Error, Root_Mean_Square_Error]],
                      columns = ['Model', 'R2 Score', 'Mean Absolute Error', 'Mean Square Error', 'Root Mean Square Error'])


results


from sklearn.ensemble import RandomForestRegressor
regressor = RandomForestRegressor(n_estimators = 100, random_state = 1)
regressor.fit(X_train, y_train)

RandomForestRegressor(random_state=1)


y_pred = regressor.predict(X_test)


R2_Score = r2_score(y_test, y_pred)
Mean_Absolute_Error = mean_absolute_error(y_test, y_pred)
Mean_Square_Error = mean_squared_error(y_test, y_pred)
Root_Mean_Square_Error = np.sqrt(mean_squared_error(y_test, y_pred))

model_results = pd.DataFrame([['Random Forest', R2_Score, Mean_Absolute_Error, Mean_Square_Error, Root_Mean_Square_Error]],
                      columns = ['Model', 'R2 Score', 'Mean Absolute Error', 'Mean Square Error', 'Root Mean Square Error'])
results = results.append(model_results, ignore_index = True)


results


from xgboost import XGBRegressor
regressor = XGBRegressor(random_state = 2)
regressor.fit(X_train, y_train)

XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
             colsample_bynode=1, colsample_bytree=1, gamma=0, gpu_id=-1,
             importance_type='gain', interaction_constraints='',
             learning_rate=0.300000012, max_delta_step=0, max_depth=6,
             min_child_weight=1, missing=nan, monotone_constraints='()',
             n_estimators=100, n_jobs=0, num_parallel_tree=1, random_state=2,
             reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,
             tree_method='exact', validate_parameters=1, verbosity=None)


y_pred = regressor.predict(X_test)


R2_Score = r2_score(y_test, y_pred)
Mean_Absolute_Error = mean_absolute_error(y_test, y_pred)
Mean_Square_Error = mean_squared_error(y_test, y_pred)
Root_Mean_Square_Error = np.sqrt(mean_squared_error(y_test, y_pred))

model_results = pd.DataFrame([['XGB Regressor', R2_Score, Mean_Absolute_Error, Mean_Square_Error, Root_Mean_Square_Error]],
                      columns = ['Model', 'R2 Score', 'Mean Absolute Error', 'Mean Square Error', 'Root Mean Square Error'])
results = results.append(model_results, ignore_index = True)


results


parameters = {"learning_rate": [0.05, 0.10, 0.15, 0.20, 0.25, 0.30], "max_depth": [3, 4, 5, 6, 8, 10, 12, 15], 
              "min_child_weight": [1, 3, 5, 7], "gamma": [0.0, 0.1, 0.2, 0.3, 0.4], "colsample_bytree": [0.3, 0.4, 0.5, 0.7]}


from sklearn.model_selection import RandomizedSearchCV
random_search = RandomizedSearchCV(estimator = regressor, param_distributions = parameters, n_iter = 5, scoring = 'r2', n_jobs = -1,
                                  cv = 10, verbose = 3)


import time

t0 = time.time()
random_search.fit(X_train, y_train)
t1 = time.time()
print("Took %0.2f Seconds" %(t1-t0))

Fitting 10 folds for each of 5 candidates, totalling 50 fits

[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.
[Parallel(n_jobs=-1)]: Done  24 tasks      | elapsed:   32.6s
[Parallel(n_jobs=-1)]: Done  50 out of  50 | elapsed:   39.8s finished

Took 41.98 Seconds


random_search.best_estimator_

XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
             colsample_bynode=1, colsample_bytree=0.5, gamma=0.0, gpu_id=-1,
             importance_type='gain', interaction_constraints='',
             learning_rate=0.05, max_delta_step=0, max_depth=12,
             min_child_weight=3, missing=nan, monotone_constraints='()',
             n_estimators=100, n_jobs=0, num_parallel_tree=1, random_state=2,
             reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,
             tree_method='exact', validate_parameters=1, verbosity=None)


random_search.best_params_

{'min_child_weight': 3,
 'max_depth': 12,
 'learning_rate': 0.05,
 'gamma': 0.0,
 'colsample_bytree': 0.5}


regressor = XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
             colsample_bynode=1, colsample_bytree=0.3, gamma=0.0, gpu_id=-1,
             importance_type='gain', interaction_constraints='',
             learning_rate=0.15, max_delta_step=0, max_depth=6,
             min_child_weight=3, monotone_constraints='()',
             n_estimators=100, n_jobs=0, num_parallel_tree=1, random_state=2,
             reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,
             tree_method='exact', validate_parameters=1, verbosity=None)


regressor.fit(X_train, y_train)

XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
             colsample_bynode=1, colsample_bytree=0.3, gamma=0.0, gpu_id=-1,
             importance_type='gain', interaction_constraints='',
             learning_rate=0.15, max_delta_step=0, max_depth=6,
             min_child_weight=3, missing=nan, monotone_constraints='()',
             n_estimators=100, n_jobs=0, num_parallel_tree=1, random_state=2,
             reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,
             tree_method='exact', validate_parameters=1, verbosity=None)


y_pred = regressor.predict(X_test)


R2_Score = r2_score(y_test, y_pred)
Mean_Absolute_Error = mean_absolute_error(y_test, y_pred)
Mean_Square_Error = mean_squared_error(y_test, y_pred)
Root_Mean_Square_Error = np.sqrt(mean_squared_error(y_test, y_pred))

model_results = pd.DataFrame([['XGB Regressor(Hyper-Parameter Tuned)', R2_Score, Mean_Absolute_Error, Mean_Square_Error, Root_Mean_Square_Error]],
                      columns = ['Model', 'R2 Score', 'Mean Absolute Error', 'Mean Square Error', 'Root Mean Square Error'])
results = results.append(model_results, ignore_index = True)


results

	POSTED_BY	UNDER_CONSTRUCTION	RERA	BHK_NO.	BHK_OR_RK	SQUARE_FT	READY_TO_MOVE	RESALE	ADDRESS	LONGITUDE	LATITUDE	TARGET(PRICE_IN_LACS)
0	Owner	0	0	2	BHK	1300.236407	1	1	Ksfc Layout,Bangalore	12.969910	77.597960	55.0
1	Dealer	0	0	2	BHK	1275.000000	1	1	Vishweshwara Nagar,Mysore	12.274538	76.644605	51.0
2	Owner	0	0	2	BHK	933.159722	1	1	Jigani,Bangalore	12.778033	77.632191	43.0
3	Owner	0	1	2	BHK	929.921143	1	1	Sector-1 Vaishali,Ghaziabad	28.642300	77.344500	62.5
4	Dealer	1	0	2	BHK	999.009247	0	1	New Town,Kolkata	22.592200	88.484911	60.5
5	Owner	0	0	3	BHK	1250.000000	1	1	South Chittoor,Kochi	10.033280	76.282571	42.0
6	Dealer	0	0	3	BHK	1495.053957	1	1	Sodala,Jaipur	26.916347	75.795600	66.5
7	Owner	0	1	3	BHK	1181.012946	1	1	Kharar,Mohali	30.740000	76.650000	52.0
8	Dealer	0	1	2	BHK	1040.000000	1	1	Bileshivale,Bangalore	13.054202	77.674002	41.6
9	Owner	0	1	2	BHK	879.120879	1	1	Chromepet,Chennai	12.951610	80.140970	36.0

	UNDER_CONSTRUCTION	RERA	BHK_NO.	SQUARE_FT	READY_TO_MOVE	RESALE	LONGITUDE	LATITUDE	TARGET(PRICE_IN_LACS)
count	29451.000000	29451.000000	29451.000000	2.945100e+04	29451.000000	29451.000000	29451.000000	29451.000000	29451.000000
mean	0.179756	0.317918	2.392279	1.980217e+04	0.820244	0.929578	21.300255	76.837695	142.898746
std	0.383991	0.465675	0.879091	1.901335e+06	0.383991	0.255861	6.205306	10.557747	656.880713
min	0.000000	0.000000	1.000000	3.000000e+00	0.000000	0.000000	-37.713008	-121.761248	0.250000
25%	0.000000	0.000000	2.000000	9.000211e+02	1.000000	1.000000	18.452663	73.798100	38.000000
50%	0.000000	0.000000	2.000000	1.175057e+03	1.000000	1.000000	20.750000	77.324137	62.000000
75%	0.000000	1.000000	3.000000	1.550688e+03	1.000000	1.000000	26.900926	77.828740	100.000000
max	1.000000	1.000000	20.000000	2.545455e+08	1.000000	1.000000	59.912884	152.962676	30000.000000

	Model	R2 Score	Mean Absolute Error	Mean Square Error	Root Mean Square Error
0	Multiple Linear Regression	0.372462	133.049526	288060.132340	536.712337
1	Random Forest	0.918399	57.487444	37457.249154	193.538754

	Model	R2 Score	Mean Absolute Error	Mean Square Error	Root Mean Square Error
0	Multiple Linear Regression	0.372462	133.049526	288060.132340	536.712337
1	Random Forest	0.918399	57.487444	37457.249154	193.538754
2	XGB Regressor	0.936873	56.221285	28977.345801	170.227336

	Model	R2 Score	Mean Absolute Error	Mean Square Error	Root Mean Square Error
0	Multiple Linear Regression	0.372462	133.049526	288060.132340	536.712337
1	Random Forest	0.918399	57.487444	37457.249154	193.538754
2	XGB Regressor	0.936873	56.221285	28977.345801	170.227336
3	XGB Regressor(Hyper-Parameter Tuned)	0.858660	63.394923	64879.348576	254.714249

Data Collection¶

Exploratory Data Analysis¶

OneHotEncoding¶

Feature Scaling¶

Splitting the dataset into Training and Testing Set**¶

Model Building¶

Multiple Linear Regression¶

Random Forest¶

XGBoost¶

Hyper-Parameter Tuning Using RandomizedSearchCV¶

Hence, we will go with the XGBoost algorithm.¶

	BHK_NO.	BHK_OR_RK
0	2	BHK
1	2	BHK
2	2	BHK
3	2	BHK
4	2	BHK
...	...	...
29446	3	BHK
29447	2	BHK
29448	2	BHK
29449	2	BHK
29450	2	BHK